index-tts

IndexTTS is an industrial-level, controllable, and highly efficient zero-shot text-to-speech (TTS) system that clones voices from a single reference audio clip. The latest IndexTTS-2.5 supports five languages (Chinese, English, Japanese, Spanish, Arabic) and features robust cross-lingual synthesis with timbre-emotion disentanglement. It offers fine-grained emotion control, speech rate adjustment, and precise pronunciation control, with native support for high-throughput production deployment via vLLM.

• Industrial-level zero-shot voice cloning
• Multilingual synthesis with phoneme-level control
• Decoupled multi-modal fine-grained emotion control
• Production-ready deployment via vLLM integration